Accelerated Gossip Algorithms for Distributed Computation

نویسندگان

  • Ming Cao
  • Daniel A. Spielman
  • Edmund M. Yeh
چکیده

We introduce a technique for accelerating the gossip algorithm of Boyd et. al. (INFOCOM 2005) for distributed averaging in a network. By employing memory in the form of a small shift-register in the computation at each node, we can speed up the algorithm’s convergence by a factor of 10. Our accelerated algorithm is inspired by the observation that the original gossip algorithm is analogous to the power method in Numerical Analysis, which can be accelerated by a shift-register based recurrence.

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Optimal Algorithms for Smooth and Strongly Convex Distributed Optimization in Networks

In this paper, we determine the optimal convergence rates for strongly convex and smooth distributed optimization in two settings: centralized and decentralized communications over a network. For centralized (i.e. master/slave) algorithms, we show that distributing Nesterov’s accelerated gradient descent is optimal and achieves a precision ε > 0 in time O(κg(1 + ∆τ) ln(1/ε)), where κg is the co...

متن کامل

A Random-Walk Model for Distributed Computation in Energy-Limited Networks

We consider two distributed algorithms to compute functions that admit flexible decomposition in terms of pairwise computations. Under these algorithms a transmitting node becomes inactive and does not transmit further messages until it is reactivated by a message reception from another node. The algorithms thereby have sequential nature and bear a close relationship to random walks. We quantif...

متن کامل

The asymptotical error of broadcast gossip averaging algorithms

In problems of estimation and control which involve a network, efficient distributed computation of averages is a key issue. This paper presents theoretical and simulation results about the accumulation of errors during the computation of averages by means of iterative “broadcast gossip” algorithms. Using martingale theory, we prove that the expectation of the accumulated error can be bounded f...

متن کامل

Concurrent and Distributed Projection through Local Interference for Wireless Sensor Networks

In this paper we use a gossip algorithm to obtain the projection of the observed signal into a subspace of lower dimension. Gossip algorithms allow distributed, fast and efficient computations on a Wireless Sensor Network and they can be properly modified to evaluate the sought projection. By combining computation coding with gossip algorithms we proposed a novel strategy that leads to importan...

متن کامل

Gossip and Random Walk Techniques for Network Coding

Ad hoc and sensor networks usually do not have a central entity for managing information spreading. Moreover, such wireless stations have limited energy and computational power. All this leads to a need for distributed and efficient algorithms for disseminating information across the network. Network coding in conjunction with random walks and gossip techniques proposes a local and distributed ...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2006